Results for “attention-matching”

Find adjacent vendors and subs (not top market primes) who fill a capability gap against a displacement target using USASpending flows and SAM entity signals. Use when user defines a teaming gap and wants vault-ready partner shortlist with citations.

Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Supports PyTorch native SDPA, flash-attn library, H100 FP8, and sliding window attention.

Expert in creating opening lines, thumbnails, and hooks that stop the scroll. Covers curiosity gaps, pattern interrupts, emotional triggers, and platform-specific hooks. Knows how to earn attention in the first 3 seconds without resorting to clickbait. Use when "hook, opening line, scroll stopper, headline, first line, attention grab, thumbnail, clickable, " mentioned.

NIH grant research skill for clinical researchers. Grill-me intake (research idea + career stage + preliminary data + environment + submission posture + known institute targets) locks down the funding strategy before any search runs. Runs a 5-facet Consensus positioning analysis (with draft Significance/Innovation language), maps the research to the right NIH institutes and study sections via RePORTER, finds NOSIs and funded overlap, and produces an editable Word document (.docx) with budget/scope-aware mechanism recommendations, submission timelines, and a mandatory program officer recommendation. Use when the user asks about research funding or makes any grant-related request (e.g., 'grants for [topic]', 'find grants for my research idea', 'what grants match my research', 'help me find NIH funding', 'grant opportunities for my research'). NIH-only scope — non-NIH funders (PCORI, DOD CDMRP, VA, foundations) are out of scope and flagged at intake.

Access real-time X/Twitter attention intelligence for Base chain tokens via the checkr API, including trending tokens, social spikes, attention/price divergence signals, and narrative summaries.

Build a library of metacognitive prompts targeting planning, monitoring, or evaluation for a specific task. Use when developing students' thinking-about-thinking during independent work.

Execute self-learning system that captures corrections during sessions and syncs them to CLAUDE.md. Use when discussing learnings, corrections, or when the user mentions remembering something. Trigger with phrases like "remember this", "don't forget", "use X not Y", or "actually...".

Diagnose and fix retention problems using the Fogg Behavior Model (B=MAP), covering motivation, ability, prompts, and tiny habits.

Diverga Memory System v7.0 - Context-persistent research support with checkpoint auto-trigger and cross-session continuity. Triggers: memory, remember, context, recall, checkpoint, decision, persist, 기억, 맥락, 세션, 체크포인트

Monitors a watchlist of Polymarket markets and surfaces meaningful shifts in price, volume, comments, or resolution proximity, with position context and optional Bankr integration.

Multi-tiered memory management (HOT/WARM/COLD) for context compaction. Invoke ONLY for explicit compaction events: post-`/compact` cleanup, MEMORY.md tier promotion, archive batch, or "trim my context". NOT for general recall (use deep-recall) or routine memory writes (use storage-router). Triggers: "compact memory", "promote to durable", "archive old context", "tier this".

Synthesize market breadth, regime, and flow signals into a unified capital commitment recommendation with exposure ceiling, bias, and action guidance.

Modify a classroom task with language scaffolds that preserve cognitive demand for EAL learners. Use when adapting existing tasks for students at different English proficiency levels.

Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Use when training/running transformers with long sequences (>512 tokens), encountering GPU memory issues with attention, or need faster inference. Supports PyTorch native SDPA, flash-attn library, H100 FP8, and sliding window attention.

当需要在众多机会中筛选出投入产出比最高的选项，以最小化风险并最大化潜在回报时

Turn accepted search evidence into an intent, competitor, and information-gain brief.

Forces unrelated concepts together to spark novel solutions and emergent properties.

Use to turn a one-off support search into a standing monitor or alert that stays useful instead of being muted within a fortnight. Trigger for "alert me when a customer mentions X", "notify the team when complaints spike", "keep watching for this", "set up a daily monitor for", turning an ad-hoc sweep into something recurring, or an existing alert nobody reads any more.

Strategic patterns for solving intractable problems through cascading approximation, self-improvement, and heterogeneous evaluation from DeepMind's AlphaGo system

Slow down interpretation from observation to action. Use when students or adults need to examine assumptions in conflict, dialogue, or inquiry.

当观察到或经历某种极端转化现象，或试图主动促成根本性观念转变时

Curating cross-agent knowledge and institutional memory: extracts patterns from agent journals into METAPATTERNS.md, detects knowledge decay, propagates best practices. Use for memory curation.

Build a weekly cadence of customer touchpoints using Opportunity Solution Trees, assumption mapping, and interview snapshots to keep product decisions grounded in evidence.

Optimize content for AI search and LLM citations across AI Overviews, ChatGPT, Perplexity, Claude, Gemini, and similar systems. Use when improving AI visibility, answer engine optimization, or citation readiness.

Diagnose and fix retention problems using behavior design (B=MAP). Use when the user mentions "users drop off", "activation rate", "onboarding friction", "retention metrics", "why users dont complete", "churn analysis", "user activation", or "aha moment". Also trigger when analyzing cohort retention curves, designing activation milestones, reducing time-to-value for new users, or investigating why users stop after their first session. Covers the Ability Chain, prompt design, and tiny behaviors that compound. For habit loops and variable rewards, see hooked-ux. For intrinsic motivation, see drive-motivation.

Mantis: Interleaved Multi-Image Instruction Tuning

Syncs key-value preferences and state across multiple agents using a shared TiDB Zero database, with optional auto-provisioning of a free ephemeral database.

Retrieve task-relevant project and global memory without loading everything. Load when the user asks what we decided, recall prior context, find memory about a feature, explain past rationale, resume a task, or check deferred ideas.

Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Use when training/running transformers with long sequences (>512 tokens), encountering GPU memory issues with attention, or need faster inference. Supports PyTorch native SDPA, flash-attn library, H100 FP8, and sliding window attention.

Use when users do not know which skill fits a task, need to search available skill catalogs, compare candidates, install a matching skill, or route work to the best existing capability.

Audits image usage across a project by comparing required image slots against available files, reporting matches, mismatches, missing assets, and broken references.

Monitor paid-ad account pacing, delivery, performance, creative fatigue, tracking, policy, and data quality across supported platforms for daily or weekly checks, anomaly review, budget pacing, post-launch verification, or campaign monitoring.

Diagnose and mitigate context degradation patterns including lost-in-middle failures, context poisoning, distraction, confusion, and clash in AI agent systems.

GLIP: Grounded Language-Image Pre-training

Handles all network operations including search, web scraping, login-required access, and social media content retrieval via a real browser CDP proxy.